CN116578890A - Intelligent factory data optimization acquisition method based on digital twinning - Google Patents

Abstract

The invention relates to the technical field of electric digital data processing, in particular to an intelligent factory data optimization acquisition method based on digital twinning, which comprises the following steps: acquiring a historical monitoring data set, current monitoring data and a historical cluster set corresponding to target plant equipment for digital twinning; performing discrete analysis processing on each history cluster in the history cluster set; determining subordinate similarity indexes between the current monitoring data and each historical cluster; determining a reference value corresponding to each parameter constituting the monitoring data; determining a target membership degree between the current monitoring data and the historical cluster; classifying the current monitoring data and the historical cluster clusters to obtain a target cluster set so as to realize the optimization of data acquisition. According to the invention, the historical monitoring data set, the current monitoring data and the historical clustering cluster are subjected to data processing, so that the technical problem of poor data acquisition effect is solved, and the optimization of data acquisition is realized.

Description

Intelligent factory data optimization acquisition method based on digital twinning

Technical Field

The invention relates to the technical field of electric digital data processing, in particular to an intelligent factory data optimization acquisition method based on digital twinning.

Background

Digital twinning means modeling an actual project by using a digital technology, thereby better managing and optimizing the project. In a factory, a plurality of parameters of equipment are monitored through sensors, so that data information for describing the running state of the current equipment is obtained, data support is provided for a plurality of working contents such as modeling, prediction, scheduling and the like in the whole production process, and safety and stability in the production process are further guaranteed. Thus, data acquisition is critical to multiple jobs in the production process. At present, when data are collected, the following modes are generally adopted: the monitoring data related to the operating state of the device is directly read.

However, when the above manner is adopted, there are often the following technical problems:

the data acquisition is carried out by directly reading the monitoring data related to the running state of the equipment, and the read monitoring data is often required to be analyzed when the engineering is managed and optimized subsequently, so that the corresponding running state of the equipment is judged, the efficiency of the engineering management and optimization is low subsequently, and therefore, the data acquisition effect is poor when the data acquisition is realized by directly reading the monitoring data related to the running state of the equipment.

Disclosure of Invention

The summary of the invention is provided to introduce a selection of concepts in a simplified form that are further described below in the detailed description. The summary of the invention is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter.

In order to solve the technical problem of poor data acquisition effect, the invention provides an intelligent factory data optimization acquisition method based on digital twinning.

The invention provides an intelligent factory data optimization acquisition method based on digital twinning, which comprises the following steps:

acquiring a historical monitoring data set corresponding to target plant equipment for digital twinning, current monitoring data and a historical cluster set corresponding to the historical monitoring data set;

performing discrete analysis processing on each history cluster in the history cluster set to obtain a discrete distribution index corresponding to the history cluster;

determining subordinate similarity indexes between the current monitoring data and each historical cluster;

determining the reference value corresponding to each parameter composing the monitoring data according to the historical monitoring data set and the current monitoring data;

Determining a target membership degree between the current monitoring data and the historical cluster according to a discrete distribution index corresponding to each historical cluster, a subordinate similarity index between the current monitoring data and the historical cluster, reference values corresponding to various parameters constituting the monitoring data, the historical cluster and the current monitoring data;

and classifying the current monitoring data and the history cluster clusters according to the target membership degree between the current monitoring data and each history cluster in the history cluster set to obtain a target cluster set so as to realize the optimization of data acquisition.

Optionally, the performing discrete analysis processing on each history cluster in the history cluster set to obtain a discrete distribution index corresponding to the history cluster includes:

determining a target position corresponding to each historical monitoring data according to normalized values corresponding to various parameters included in each historical monitoring data;

screening a core monitoring data set from the historical cluster according to a target position corresponding to the historical monitoring data, wherein the number of the historical monitoring data in a range surrounded by a preset radius corresponding to the core monitoring data is larger than the preset number;

Determining an aggregation position corresponding to the history cluster according to the target position corresponding to each core monitoring data in the core monitoring data set;

the distance between the target position corresponding to each history monitoring data in the history cluster and the gathering position is used as the target distance corresponding to the history monitoring data, and a target distance set corresponding to the history cluster is obtained;

and determining standard deviations of all target distances in the target distance set corresponding to the history cluster as discrete distribution indexes corresponding to the history cluster.

Optionally, the determining a subordinate similarity index between the current monitored data and each historical cluster includes:

determining a current position corresponding to the current monitoring data according to normalized values corresponding to various parameters included in the current monitoring data;

connecting the current position with the aggregation position to obtain a reference line segment between the current monitoring data and the historical cluster;

determining a vertical line between a target position corresponding to each historical monitoring data in the historical cluster and the reference line segment as a target vertical line corresponding to the historical monitoring data;

Screening historical monitoring data of the intersection of the target vertical line and the reference line segment from the historical cluster as reference monitoring data to obtain a reference monitoring data set between the current monitoring data and the historical cluster;

connecting the gathering position with a target position corresponding to each piece of reference monitoring data in the reference monitoring data set to obtain a target line segment corresponding to the reference monitoring data;

determining a first subordinate index between the current monitoring data and the historical cluster according to the reference line segment and a target line segment corresponding to each reference monitoring data in the reference monitoring data set;

determining the number of reference monitoring data in the reference monitoring data set, and the ratio of the number of historical monitoring data in the historical cluster as a second subordinate index between the current monitoring data and the historical cluster;

and determining subordinate similarity indexes between the current monitoring data and the history cluster according to the first subordinate indexes and the second subordinate indexes, wherein the first subordinate indexes and the second subordinate indexes are positively correlated with the subordinate similarity indexes.

Optionally, the determining, according to the reference line segment and the target line segment corresponding to each reference monitoring data in the reference monitoring data set, a first subordinate indicator between the current monitoring data and the historical cluster includes:

determining the projection of a target line segment corresponding to each reference monitoring data in the reference monitoring data set on the reference line segment as the target projection corresponding to the reference monitoring data;

and determining a first subordinate index between the current monitoring data and the historical cluster according to target projections corresponding to each reference monitoring data in the reference monitoring data set, wherein the target projections and the first subordinate index are positively correlated.

Optionally, the determining, according to the historical monitoring data set and the current monitoring data, a reference value corresponding to each parameter that constitutes the monitoring data includes:

screening a candidate monitoring data set from a set formed by the current monitoring data and the historical monitoring data set according to the corresponding acquisition time of the historical monitoring data;

determining a normalized value corresponding to the parameter included in each candidate monitoring data in the candidate monitoring data set as a target parameter value corresponding to the parameter included in the candidate monitoring data, so as to obtain a target parameter value set corresponding to the parameter;

Determining the average value of all target parameter values in the target parameter value set as a representative parameter value corresponding to the parameter;

determining the absolute value of the difference value between the target parameter value corresponding to the parameter and the representative parameter value included in each candidate monitoring data in the candidate monitoring data set as a parameter difference index corresponding to the parameter included in the candidate monitoring data;

determining a difference value between the current time and the acquisition time corresponding to each candidate monitoring data in the candidate monitoring data set as a time span corresponding to the candidate monitoring data;

and determining a reference value corresponding to the parameter according to the parameter difference index corresponding to the parameter and the time span corresponding to each candidate monitoring data included in each candidate monitoring data in the candidate monitoring data set, wherein the time span and the parameter difference index are in negative correlation with the reference value.

Optionally, the determining the target membership between the current monitoring data and the historical cluster includes:

determining a first membership degree between the current monitoring data and the historical cluster according to the discrete distribution index and the subordinate similarity index, wherein the discrete distribution index is in negative correlation with the first membership degree, and the subordinate similarity index is in positive correlation with the first membership degree;

Determining a second membership degree of the current monitoring data and the historical monitoring data corresponding to the parameters according to the reference value corresponding to each parameter composing the monitoring data, the normalized value corresponding to the parameter included in each historical monitoring data in the historical cluster and the normalized value corresponding to the parameter included in the current monitoring data;

determining a third membership degree between the current monitoring data and the history clustering cluster according to the second membership degree of each history monitoring data in the current monitoring data and the history clustering cluster corresponding to each parameter, wherein the second membership degree and the third membership degree are positively correlated;

and determining the target membership degree between the current monitoring data and the history cluster according to the first membership degree and the third membership degree, wherein the first membership degree and the third membership degree are positively correlated with the target membership degree.

Optionally, the determining, according to the reference value corresponding to each parameter that constitutes the monitoring data, the normalized value corresponding to the parameter that is included in each historical monitoring data in the historical cluster, and the normalized value corresponding to the parameter that is included in the current monitoring data, the second membership degree of the current monitoring data and the historical monitoring data corresponding to the parameter includes:

Determining the absolute value of the difference value between the normalized value corresponding to the parameter included in the historical monitoring data and the normalized value corresponding to the parameter included in the current monitoring data as the data difference between the current monitoring data and the historical monitoring data corresponding to the parameter;

and determining a second membership degree of the current monitoring data and the historical monitoring data corresponding to the parameters according to the reference value and the data difference corresponding to the parameters composing the monitoring data, wherein the reference value and the second membership degree are positively correlated, and the data difference and the second membership degree are negatively correlated.

Optionally, the classifying the current monitoring data and the history cluster clusters according to the target membership degree between the current monitoring data and each history cluster in the history cluster set to obtain a target cluster set includes:

screening a history cluster with the maximum target membership degree from the history cluster set to be used as a candidate cluster;

determining historical clusters except the candidate clusters in the historical cluster set as target clusters;

adding the current monitoring data to the candidate cluster, and determining the added candidate cluster as a target cluster;

And forming a target cluster set by all the obtained target clusters.

The invention has the following beneficial effects:

according to the intelligent factory data optimization collection method based on digital twinning, the data processing is carried out on the historical monitoring data set, the current monitoring data and the historical clustering cluster, so that the technical problem of poor data collection effect is solved, and the data collection optimization is realized. Firstly, a historical monitoring data set, current monitoring data and a historical cluster set are obtained, so that the subsequent classification of the current monitoring data can be facilitated, and the subsequent judgment of the running state of equipment corresponding to each historical monitoring data and the current monitoring data in the historical monitoring data set can be facilitated. Then, each history cluster in the history cluster set is subjected to discrete analysis processing, so that the accuracy of determining the discrete distribution index corresponding to the history cluster can be improved. And then, quantifying subordinate similarity indexes between the current monitoring data and each historical cluster, so that the running state of the equipment corresponding to the current monitoring data can be conveniently and subsequently judged. Continuing, the historical monitoring data set and the current monitoring data are comprehensively considered, so that the accuracy of reference value determination corresponding to each parameter composing the monitoring data can be improved. And then, comprehensively considering the discrete distribution index corresponding to each historical cluster, the subordinate similarity index between the current monitoring data and each historical cluster, the reference value corresponding to various parameters composing the monitoring data, and each historical cluster and the current monitoring data, so that the accuracy of determining the target membership degree can be improved. Finally, comprehensively considering the target membership degree between the current monitoring data and each historical cluster in the historical cluster set, classifying the current monitoring data and the historical cluster set, and obtaining the target cluster to which each monitoring data belongs, namely, realizing the state division of the current monitoring data in the data acquisition process, so that the state division of the current monitoring data is not needed when the engineering is managed and optimized in the follow-up process, the efficiency of managing and optimizing the engineering in the follow-up process can be improved, and the target cluster set is determined, so that the follow-up analysis and treatment can be facilitated.

Drawings

In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.

FIG. 1 is a flow chart of the intelligent plant data optimization acquisition method based on digital twinning.

Detailed Description

In order to further describe the technical means and effects adopted by the present invention to achieve the preset purpose, the following detailed description is given below of the specific implementation, structure, features and effects of the technical solution according to the present invention with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

The invention provides an intelligent factory data optimization acquisition method based on digital twinning, which comprises the following steps:

acquiring a historical monitoring data set corresponding to the target plant equipment for digital twinning, current monitoring data and a historical cluster set corresponding to the historical monitoring data set;

performing discrete analysis processing on each history cluster in the history cluster set to obtain a discrete distribution index corresponding to the history cluster;

determining subordinate similarity indexes between the current monitoring data and each historical cluster;

determining the reference value corresponding to each parameter composing the monitoring data according to the historical monitoring data set and the current monitoring data;

determining target membership between the current monitoring data and the historical cluster according to discrete distribution indexes corresponding to each historical cluster, subordinate similarity indexes between the current monitoring data and the historical clusters, reference values corresponding to various parameters constituting the monitoring data, the historical cluster and the current monitoring data;

classifying the current monitoring data and the historical cluster clusters according to the target membership degree among each historical cluster in the current monitoring data and the historical cluster sets to obtain a target cluster set so as to realize the optimization of data acquisition.

The following detailed development of each step is performed:

referring to FIG. 1, a flow of some embodiments of a digital twinning-based intelligent plant data optimization acquisition method in accordance with the present invention is shown. The intelligent factory data optimization acquisition method based on digital twinning comprises the following steps:

step S1, a historical monitoring data set corresponding to the target plant equipment for digital twinning, current monitoring data and a historical cluster set corresponding to the historical monitoring data set are obtained.

In some embodiments, a set of historical monitoring data corresponding to a target plant device for digital twinning, current monitoring data, and a set of historical cluster clusters corresponding to the set of historical monitoring data may be obtained.

Wherein the target plant device may be a device in the plant that participates in digital twinning. The historical monitoring data in the historical monitoring data set may be monitoring data collected over a historical period of time that is related to the operational status of the target plant device. The current monitoring data may be monitoring data collected at the current time that is related to the operational status of the target plant device. The history period is a period before the current time. The monitoring data may include a plurality of parameters related to the operating state of the target plant equipment. The monitoring data may be data composed of normalized parameter values. The parameter value may be a value corresponding to the parameter. For example, if the target plant is an air conditioner, the plurality of parameters related to the operating state of the air conditioner may include: refrigeration capacity, energy efficiency ratio, noise and heating capacity, the monitoring data may include: the value corresponding to the normalized refrigerating capacity, the value corresponding to the normalized energy efficiency ratio, the value corresponding to the normalized noise and the value corresponding to the normalized heating capacity are obtained. The historical cluster in the historical cluster set may be a cluster obtained by clustering historical monitoring data in the historical monitoring data set.

It should be noted that digital twinning means modeling an actual engineering by using digital technology, so as to better manage and optimize the engineering. In a factory, a plurality of parameters of equipment are monitored through sensors, so that data information for describing the running state of the equipment is obtained, and data support is provided for a plurality of working contents such as modeling, prediction, scheduling and the like in the whole production process, so that safety and stability in the production process are guaranteed. The historical monitoring data set, the current monitoring data and the historical cluster set are acquired, so that the subsequent classification of the current monitoring data can be facilitated, and the subsequent judgment of the running state of each historical monitoring data in the historical monitoring data set and the corresponding equipment of the current monitoring data can be facilitated.

As an example, this step may include the steps of:

the method comprises the steps of firstly, collecting values corresponding to a plurality of parameters related to the running state of target factory equipment at each preset time in a historical time period through a sensor, and combining the values of the plurality of parameters collected at each preset time into initial data corresponding to each preset time.

The preset time may be a preset time. The time period between adjacent preset times may be a preset time period. The preset time period may be a preset time period. For example, the preset duration may be 1 second. The initial data corresponding to the preset time may include: a plurality of parameter values collected at the preset moment.

For example, if the preset duration is 1 second, initial data may be collected by the sensor at 1 second intervals over the historical period.

And secondly, acquiring values corresponding to a plurality of parameters related to the running state of the target factory equipment at the current moment through a sensor, and combining the plurality of parameter values acquired at the current moment into initial data corresponding to the current moment.

The duration between the current time and the ending time included in the historical time period may be a preset duration.

And thirdly, normalizing all parameter values included in the obtained initial data to achieve normalization of the initial data, and taking the normalized initial data as monitoring data.

For example, taking any parameter as an example, normalizing the values corresponding to the parameter included in all the obtained initial data may include the following substeps:

and a first sub-step of screening out the values corresponding to the parameters from all initial data in the historical time period corresponding to the acquisition time as candidate values to obtain a candidate value set.

For example, if the parameter is speed, the candidate set of values may include: the corresponding acquisition time includes all the velocity values included in the initial data during the historical time period.

And a second sub-step of screening out the largest candidate value from the candidate value set as a first value, and screening out the smallest candidate value from the candidate value set as a second value.

And a third sub-step of determining the average value of the first numerical value and the second numerical value as a target average value.

And a fourth sub-step of determining the difference between the first value and the target mean value as a first step size.

And a fifth substep, determining the product of the first step length and the preset multiple as a second step length.

The preset multiple may be a preset multiple. For example, the preset multiple may be 1.3.

And a sixth substep, determining the sum of the target average value and the second step length as a target maximum value.

And a seventh substep, determining the difference between the target average value and the second step length as a target minimum value.

And eighth substep, screening out the value corresponding to the parameter from all the obtained initial data to be used as the value to be determined, and obtaining a set of values to be determined.

And a ninth substep, taking the target maximum value as the maximum value, taking the target minimum value as the minimum value, normalizing each undetermined value in the undetermined value set through maximum and minimum normalization, and realizing normalization of the value corresponding to the parameter.

It should be noted that, after normalization is performed on a plurality of parameter values included in the initial data, normalization of the initial data may be implemented. Second, in some cases, some parameters may exhibit abrupt values, resulting in them exceeding the range bounded by the corresponding initial maximum and minimum values, and thus the range may be extended to some extent, for example, by a factor of 1.3 in the step size from the center point to the end points of the range. Wherein the range center point may be characterized by a target mean. The endpoints herein may be corresponding maxima or minima.

And fourthly, taking the corresponding monitoring data with the acquisition time being the current moment as the current monitoring data, taking the monitoring data with the corresponding acquisition time in the historical time period as the historical monitoring data, and combining all the obtained historical monitoring data into the historical monitoring data.

And fifthly, clustering historical monitoring data in the historical monitoring data set through a DBSCAN (Density-Based Spatial Clustering of Applications with Noise based clustering) algorithm, taking the obtained cluster as a historical cluster, and forming a historical cluster set from all the obtained historical clusters.

It should be noted that, the running states of the devices corresponding to the historical monitoring data in the same historical cluster are often similar. The historical cluster can be a cluster which is clustered in advance, and the historical cluster can also be a cluster which judges the running state of the corresponding equipment in advance.

And S2, performing discrete analysis processing on each history cluster in the history cluster set to obtain a discrete distribution index corresponding to the history cluster.

In some embodiments, discrete analysis processing may be performed on each history cluster in the history cluster set to obtain a discrete distribution index corresponding to the history cluster.

It should be noted that, discrete analysis processing is performed on each history cluster in the history cluster set, so that accuracy of determining the discrete distribution index corresponding to the history cluster can be improved.

As an example, this step may include the steps of:

first, determining a target position corresponding to each historical monitoring data according to normalized values corresponding to various parameters included in each historical monitoring data.

The normalized value may be a value obtained by normalization. For example, if a certain parameter is the cooling capacity, the normalized value corresponding to the cooling capacity may be a value corresponding to the normalized cooling capacity.

For example, normalized values corresponding to various parameters included in the historical monitoring data may be combined into coordinates, and the coordinates may be used to characterize the target location corresponding to the historical monitoring data. If the plurality of parameters included in the historical monitoring data include: the refrigerating capacity, the energy efficiency ratio, the noise and the heating capacity are recorded as a first historical moment at the acquisition moment corresponding to the historical monitoring data, and then the target position corresponding to the historical monitoring data can be a coordinate (a normalized value corresponding to the refrigerating capacity acquired at the first historical moment, a normalized value corresponding to the energy efficiency ratio acquired at the first historical moment, a normalized value corresponding to the noise acquired at the first historical moment and a normalized value corresponding to the heating capacity acquired at the first historical moment).

It should be noted that, since the dimensions of various parameters included in the history monitoring data are often different, the target positions corresponding to the history monitoring data are determined by adopting normalized values corresponding to the various parameters, so that the influence of the different dimensions can be eliminated.

And step two, screening out a core monitoring data set from the history cluster according to the target position corresponding to the history monitoring data.

The number of the historical monitoring data within the range surrounded by the preset radius corresponding to the core monitoring data can be larger than the preset number. The preset radius may be a preset radius. The preset radius may be a radius set when the historical monitoring data is clustered through a DBSCAN algorithm. For example, the preset radius may be 2. The preset number may be a preset number. The preset number may be a minimum data amount required for forming a cluster set when the historical monitoring data is clustered through a DBSCAN algorithm. For example, the preset number may be 10. The core monitoring data in the core monitoring data set may be core points in a DBSCAN algorithm.

And thirdly, determining the aggregation position corresponding to the history cluster according to the target position corresponding to each core monitoring data in the core monitoring data set.

For example, the center of the target position corresponding to each core monitoring data in the core monitoring data set corresponding to the history cluster may be determined as the aggregation position corresponding to the history cluster.

For example, if the set of core monitoring data corresponding to a certain historical cluster includes 3 core monitoring data, and the target positions corresponding to the 3 core monitoring data may be (0.1,0.2,0.3), (0.4,0.5,0.6) and (0.7,0.8,0.9), respectively, because，/>，/>The coordinates corresponding to the aggregate location corresponding to the historical cluster may be (0.4,0.5,0.6).

And step four, taking the distance between the target position corresponding to each history monitoring data in the history cluster and the gathering position as the target distance corresponding to the history monitoring data to obtain a target distance set corresponding to the history cluster.

For example, the distance between the target position corresponding to the history monitoring data and the aggregation position corresponding to the history cluster in which the history monitoring data is located may be used as the target distance corresponding to the history monitoring data.

And fifthly, determining standard deviations of all target distances in the target distance set corresponding to the history cluster as discrete distribution indexes corresponding to the history cluster.

For example, the formula corresponding to the discrete distribution index corresponding to the history cluster may be determined as follows:

wherein,,is the +.>Discrete distribution indexes corresponding to the historical clusters. />Is the firstThe number of history monitoring data in each history cluster. />Is->The>Target distances corresponding to the historical monitoring data; i.e. < ->Target position corresponding to historical monitoring data and +.>The distance between the aggregation positions corresponding to the historical clusters. />Is->And the average value of the target distances corresponding to all the historical monitoring data in the historical cluster. />Is the serial number of the history cluster in the history cluster set. />Is->Sequence numbers of historical monitoring data in the historical clusters. />Is->Square of (d).

When the following is performedThe greater the case, the more +.>The more chaotic the target positions corresponding to the historical monitoring data in the historical cluster are, the description of the +.>The lower the similarity between the history monitored data in the history clusters, the more often the description +. >The worse the clustering effect of the historical clusters is, the less the current monitoring data should be divided into the firstAnd (5) clustering the histories.

And S3, determining subordinate similarity indexes between the current monitoring data and each historical cluster.

In some embodiments, a subordinate similarity index between the current monitored data and each historical cluster may be determined.

It should be noted that, the subordinate similarity index between the current monitoring data and each historical cluster is quantized, so that the running state of the device corresponding to the current monitoring data can be conveniently and subsequently judged.

As an example, this step may include the steps of:

the first step, determining the current position corresponding to the current monitoring data according to the normalized values corresponding to various parameters included in the current monitoring data.

For example, normalized values corresponding to various parameters included in the current monitoring data may be combined into coordinates, and the coordinates may be used to characterize the current position corresponding to the current monitoring data. If the current monitoring data includes a plurality of parameters including: the current position corresponding to the current monitoring data can be coordinates (normalized value corresponding to the refrigerating capacity collected at the current moment, normalized value corresponding to the energy efficiency ratio collected at the current moment, normalized value corresponding to the noise collected at the current moment, and normalized value corresponding to the heating capacity collected at the current moment).

And a second step of connecting the current position with the gathering position to obtain a reference line segment between the current monitoring data and the history clustering cluster.

For example, the current monitoring data and the firstThe reference line segments between the historical clusters may be: connect the current location with->And obtaining line segments at the gathering positions corresponding to the historical clusters.

And thirdly, determining the vertical line between the target position corresponding to each history monitoring data in the history cluster and the reference line segment as the target vertical line corresponding to the history monitoring data.

And step four, screening historical monitoring data of the intersection of the target vertical line and the reference line segment from the historical cluster, and taking the historical monitoring data as reference monitoring data to obtain a reference monitoring data set between the current monitoring data and the historical cluster.

And fifthly, connecting the gathering position with a target position corresponding to each piece of reference monitoring data in the reference monitoring data set to obtain a target line segment corresponding to the reference monitoring data.

And a sixth step of determining a first subordinate index between the current monitoring data and the historical cluster according to the reference line segment and a target line segment corresponding to each reference monitoring data in the reference monitoring data set.

For example, determining a first subordinate indicator between the current monitored data and the historical cluster according to the reference line segment and a target line segment corresponding to each reference monitored data in the set of reference monitored data may include the sub-steps of:

and a first sub-step of determining the projection of the target line segment corresponding to each piece of reference monitoring data in the reference monitoring data set on the reference line segment as the target projection corresponding to the reference monitoring data.

And a second sub-step of determining a first subordinate index between the current monitoring data and the history cluster according to the target projection corresponding to each reference monitoring data in the reference monitoring data set.

Wherein the target projection may be positively correlated with the first slave index.

And seventh, determining the duty ratio of the number of the reference monitoring data in the reference monitoring data set and the number of the historical monitoring data in the historical cluster as a second subordinate index between the current monitoring data and the historical cluster.

And eighth step, determining subordinate similarity indexes between the current monitoring data and the history cluster according to the first subordinate indexes and the second subordinate indexes.

The first subordinate index and the second subordinate index may both be positively correlated with the subordinate similarity index.

For example, the formula for determining subordinate similarity index correspondence between the current monitored data and the historical cluster may be:

wherein,,is the first ∈of the current monitored data and historical cluster set>Subordinate similarity index between historical clusters. />Is->The number of history monitoring data in each history cluster. />Is the current monitoring data and +.>The number of reference monitor data in the reference monitor data set between the historical clusters. />Is the current monitoring data and the firstIndividual history clusteringIn the reference monitoring data set between clusters, +.>And projecting the target corresponding to the reference monitoring data. />Is the current monitoring data and +.>In the reference monitoring data set among the historical clusters +.>The distance of the target line segment corresponding to the reference monitoring data. />Is the current monitoring data and +.>In the reference monitoring data set among the historical clusters +.>Included angles between target line segments corresponding to the reference monitoring data and a first line segment, wherein the first line segment is the current monitoring data and +.>Reference line segments between the historical clusters. / >Is->Is a cosine of (c). />Is the current monitoring data and +.>A first subordinate index between the historical clusters. />And->And shows positive correlation. />Is the current monitoring data and +.>And a second subordinate index between the history clusters. />And->All are in charge of>And shows positive correlation. />Is the serial number of the history cluster in the history cluster set. />Is the current monitoring data and +.>Sequence numbers of reference monitoring data in the reference monitoring data sets among the historical clusters.

When the following is performedThe greater the case, the more +.>The larger the projection of the reference monitor data relative to the current monitor data, the more often it is explained +.>The more relevant the reference monitoring data and the current monitoring data areStrong. When->The larger the reference monitored data similar to the current monitored data is, the more the reference monitored data is relatively, the more the current monitored data is likely to be divided into +.>And (5) clustering the histories. Thus, when->The larger the current monitoring data, the more the device operation state corresponding to the current monitoring data is, the more the current monitoring data is>The more similar the running states of the devices corresponding to the historical clusters are, the more the current monitoring data should be divided into the (th)>And (5) clustering the histories.

And S4, determining the reference value corresponding to each parameter constituting the monitoring data according to the historical monitoring data set and the current monitoring data.

In some embodiments, a reference value corresponding to each parameter that makes up the monitoring data may be determined from the set of historical monitoring data and the current monitoring data.

It should be noted that, by comprehensively considering the historical monitoring data set and the current monitoring data, the accuracy of determining the reference value corresponding to each parameter constituting the monitoring data can be improved.

As an example, this step may include the steps of:

first, screening a candidate monitoring data set from a set formed by the current monitoring data and the historical monitoring data set according to the collection time corresponding to the historical monitoring data.

Wherein the set of current monitoring data and historical monitoring data sets may include: current monitoring data and the historical monitoring data set.

For example, a preset number of monitoring data closest to the current time may be selected from a set of current monitoring data and a set of historical monitoring data, and the set of preset number of monitoring data is used as a candidate monitoring data set. Wherein the preset number may be a preset number. For example, the preset number may be 30.

For example, if the preset number is 3, the current time is 2023, 07, 04, 14, 00 minutes and 04 seconds, and the historical monitoring data set includes: the first historical monitoring data, the second historical monitoring data and the third historical monitoring data, the acquisition time corresponding to the current monitoring data is 2023, 07, month, 04, day 14, 00 minutes and 04 seconds, the acquisition time corresponding to the third historical monitoring data is 2023, 07, month, 04, day 14, 00 minutes and 03 seconds, the acquisition time corresponding to the second historical monitoring data is 2023, 07, month, 04, day 14, 00 minutes and 02 seconds, the acquisition time corresponding to the first historical monitoring data is 2023, 07, month, 04, day 14, 00 minutes and 01 seconds, then the candidate monitoring data set may include: the second historical monitoring data, the third historical monitoring data and the current monitoring data.

And a second step of determining the normalized value corresponding to the parameter included in each candidate monitoring data in the candidate monitoring data set as a target parameter value corresponding to the parameter included in the candidate monitoring data, thereby obtaining a target parameter value set corresponding to the parameter.

The target parameter value set corresponding to a certain parameter may include: all candidate monitoring data comprise target parameter values corresponding to the parameters.

For example, if a certain parameter is speed, the target parameter value corresponding to the speed included in the candidate monitoring data may be a normalized value corresponding to the speed included in the candidate monitoring data. The set of speed-corresponding target parameter values may include: all candidate monitoring data comprise target parameter values corresponding to the speeds.

And thirdly, determining the average value of all the target parameter values in the target parameter value set as a representative parameter value corresponding to the parameter.

And a fourth step of determining an absolute value of a difference between a target parameter value corresponding to the parameter included in each candidate monitoring data set and the representative parameter value as a parameter difference index corresponding to the parameter included in the candidate monitoring data set.

And fifthly, determining a difference value between the current time and the acquisition time corresponding to each candidate monitoring data in the candidate monitoring data set as a time span corresponding to the candidate monitoring data.

And sixthly, determining the reference value corresponding to the parameter according to the parameter difference index corresponding to the parameter and the time span corresponding to each candidate monitoring data included in each candidate monitoring data in the candidate monitoring data set.

Wherein, the time span and the parameter difference index can be in negative correlation with reference value.

For example, the formula for determining the reference value for each parameter constituting the monitoring data may be:

wherein,,is the +.>And the reference value corresponding to the parameter. />Is the number of candidate monitoring data in the candidate monitoring data set. />Is the +.f. in the composition candidate monitoring data set>The candidate monitoring data comprises +.>First value corresponding to seed parameterAnd (5) an index. />Is a natural constant +.>To the power. />Is thatAbsolute values of (2), i.e. make up the (th) of the candidate monitoring data set>The candidate monitoring data comprises +.>And a parameter difference index corresponding to the parameters. />Is the +.f. in the composition candidate monitoring data set >The candidate monitoring data comprises +.>Normalized values corresponding to the seed parameters, i.e. +.>The candidate monitoring data comprises +.>And a target parameter value corresponding to the parameter. The first item of monitoring data included in each of the set of candidate monitoring data>The parameters are the same parameters. />Is the first to compose the monitoring data/>Representative parameter values corresponding to the seed parameters; i.e. the +.th of all candidate monitoring data included in the composition candidate monitoring data set>Average value of normalized values corresponding to seed parameters; i.e. < ->The average value of all target parameter values in the target parameter value set corresponding to the parameters. />Is->The time span corresponding to the candidate monitoring data, i.e. the current time and +.>Differences between acquisition times corresponding to the candidate monitoring data. />Is->Square of (d). />And->All can be combined with->And has negative correlation. />Is the sequence number of the parameter that constitutes the monitoring data. />Is a candidate monitoringSequence numbers of candidate monitoring data in the data set. />Can be in the range of [0.5,1 ]]。

When the following is performedThe greater the case, the more +.>The candidate monitoring data comprises +.>The larger the fluctuation of the normalized value corresponding to the parameter. When- >The smaller the time, the more +.>The closer the acquisition time corresponding to the candidate monitoring data is to the current moment, the description of the +.>The smaller the time span between the candidate monitoring data and the current monitoring data, the more often the explanation +.>The higher the sensitivity of the candidate monitoring data to the current monitoring data, the more often it is explained +.>The higher the confidence that the candidate monitoring data is for the current monitoring data. Thus, when->The smaller the time, the more +.>The more the fluctuation degree of the normalized value corresponding to the seed parameterLarge, it is often explained that each candidate monitoring data includes +.>The more disturbed the variation of the normalized values corresponding to the seed parameters, the description of +.>The more unstable the normalized value corresponding to the seed parameter, the more often the explanation is +.>The lower the reference value of a parameter relative to the current monitored data.

And S5, determining the target membership degree between the current monitoring data and the historical clustering clusters according to the discrete distribution index corresponding to each historical clustering cluster, the subordinate similarity index between the current monitoring data and the historical clustering clusters, the reference value corresponding to various parameters constituting the monitoring data, the historical clustering clusters and the current monitoring data.

In some embodiments, the target membership degree between the current monitoring data and the historical cluster may be determined according to a discrete distribution index corresponding to each historical cluster, a subordinate similarity index between the current monitoring data and the historical cluster, a reference value corresponding to various parameters constituting the monitoring data, the historical cluster and the current monitoring data.

It should be noted that, the accuracy of determining the target membership degree can be improved by comprehensively considering the discrete distribution index corresponding to each history cluster, the subordinate similarity index between the current monitoring data and each history cluster, the reference value corresponding to various parameters constituting the monitoring data, and each history cluster and the current monitoring data.

As an example, this step may include the steps of:

and a first step of determining a first membership degree between the current monitoring data and the history cluster according to the discrete distribution index and the subordinate similarity index.

Wherein the discrete distribution index may be inversely related to the first membership degree. The subordinate similarity measure may be positively correlated with the first membership.

And a second step of determining a second membership degree of the current monitoring data and the historical monitoring data corresponding to the parameters according to the reference value corresponding to each parameter composing the monitoring data, the normalized value corresponding to the parameter included in each historical monitoring data in the historical cluster and the normalized value corresponding to the parameter included in the current monitoring data.

For example, determining the second membership degree of the current monitoring data and the historical monitoring data corresponding to the parameters according to the reference value corresponding to each parameter constituting the monitoring data, the normalized value corresponding to the parameter included in each historical monitoring data in the historical cluster, and the normalized value corresponding to the parameter included in the current monitoring data may include the following substeps:

and a first sub-step of determining an absolute value of a difference between a normalized value corresponding to the parameter included in the history monitoring data and a normalized value corresponding to the parameter included in the current monitoring data as a data difference between the current monitoring data and the history monitoring data corresponding to the parameter.

And a second sub-step of determining a second membership degree of the current monitoring data and the historical monitoring data corresponding to the parameters according to the reference value and the data difference corresponding to the parameters composing the monitoring data.

Wherein the reference value may be positively correlated with the second degree of membership. The data difference may be inversely related to the second membership.

And thirdly, determining a third membership degree between the current monitoring data and the history cluster according to the second membership degree of each history monitoring data in the current monitoring data and the history cluster corresponding to each parameter.

Wherein the second degree of membership may be positively correlated with the third degree of membership.

And step four, determining the target membership degree between the current monitoring data and the history cluster according to the first membership degree and the third membership degree.

Wherein, the first membership degree and the third membership degree can be positively correlated with the target membership degree.

For example, the formula for determining the target membership correspondence between the current monitored data and the historical cluster may be:

wherein,,is the first in the current monitoring data and history cluster setmTarget membership between historical clusters. />Is->Discrete distribution indexes corresponding to the historical clusters. />Is the current monitoring data and +.>Subordinate similarity index between historical clusters. />Is the current monitoring data and +.>And a third degree of membership between the historical clusters. />Is->The number of history monitoring data in each history cluster. />Is the total number of parameters that make up the monitored data.Is the current monitoring data and +.>The>Historical monitoring data at the first place of the composition monitoring dataThe data difference corresponding to the seed parameters, i.e. the +.>Normalized value and the first of the parameters The history monitoring data comprises +.>Absolute value of difference of normalized values corresponding to the parameters. />Is the +.>And the reference value corresponding to the parameter. />Is a natural constant +.>To the power. />Is a natural constant +.>To the power. />Is the current monitoring data and +.>The>Historical monitoring data in the +.>And a second membership corresponding to the parameter. />And (3) withAnd has negative correlation. />And->And shows positive correlation. />Is the current monitoring data and +.>A first degree of membership between the historical clusters. />And->And has negative correlation. />And->And shows positive correlation. />And->And shows positive correlation.And->All are in charge of>And shows positive correlation. />Is the serial number of the history cluster in the history cluster set. />Is->Sequence numbers of historical monitoring data in the historical clusters. />Is the sequence number of the parameter that constitutes the monitoring data.

When the following is performedThe larger the tends to explain the firstiThe higher the reference value of a parameter relative to the current monitored data. When->The smaller the time, the more current monitoring data and +.>The>Historical monitoring data at the first place of the composition monitoring dataiThe more similar the corresponding values of the parameters. So when->The larger the current monitoring data, the more the device operation state corresponding to the current monitoring data is, the more the current monitoring data is >The more similar the running states of the devices corresponding to the historical clusters are, the more the current monitoring data should be divided into the (th)>And (5) clustering the histories. When->The larger the current monitoring data is, the more the current monitoring data should be divided into +.>And (5) clustering the histories. When->The greater the case, the more +.>The more chaotic the target positions corresponding to the historical monitoring data in the historical cluster are, the description of the +.>The lower the similarity between the history monitored data in the history clusters, the more often the description +.>The worse the clustering effect of the historical clusters is, the less the current monitoring data should be divided into the firstAnd (5) clustering the histories. Thus, when->The larger the current monitoring data is, the more the current monitoring data should be divided into +.>And (5) clustering the histories.

And S6, classifying the current monitoring data and the historical cluster clusters according to the target membership degree between the current monitoring data and each historical cluster in the historical cluster set to obtain a target cluster set so as to realize the optimization of data acquisition.

In some embodiments, the current monitoring data and the history cluster clusters may be classified according to a target membership degree between the current monitoring data and each history cluster in the history cluster set, so as to obtain a target cluster set, so as to implement optimization of data collection.

It should be noted that, comprehensively considering the target membership degree between the current monitoring data and each history cluster in the history cluster set, classifying the current monitoring data and the history cluster set, and obtaining the target cluster to which each monitoring data belongs. The method and the device can realize the state division of the current monitoring data in the data acquisition process, so that the state division of the current monitoring data is not needed when the engineering is managed and optimized in the follow-up process, and the efficiency of the management and optimization of the engineering in the follow-up process can be improved, and therefore, the optimization of the data acquisition is realized.

As an example, this step may include the steps of:

first, screening out the history cluster with the largest target membership degree from the history cluster, and taking the history cluster as a candidate cluster.

And secondly, determining historical clusters except the candidate clusters in the historical cluster set as target clusters.

And thirdly, adding the current monitoring data to the candidate cluster, and determining the added candidate cluster as a target cluster.

Fourth, all the obtained target clusters are formed into a target cluster set.

It should be noted that, in general, the current monitoring data often belongs to a certain history cluster, so that the current monitoring data can be directly divided into the history clusters corresponding to the largest target membership. However, the current monitoring data may not be similar to each history cluster, and may not belong to any history cluster at this time, and may be marked as discrete points at this time, i.e. if the maximum target membership in all target membership is smaller than a preset membership threshold, the current monitoring data is marked as discrete points. The membership threshold may be a preset threshold. For example, the membership threshold may be 0.5. When the maximum target membership is greater than or equal to the membership threshold, processing may be performed as an example included in step S6.

In summary, the invention quantifies the target membership degree between the current monitoring data and each history cluster in the history cluster set, classifies the current monitoring data and the history cluster set, and can obtain the target cluster to which each monitoring data belongs, namely, the state division of the current monitoring data can be realized in the data acquisition process, so that the state division of the current monitoring data is not needed when the engineering is managed and optimized in the follow-up process, the follow-up efficiency of managing and optimizing the engineering can be improved, and the target cluster set is determined, so that the follow-up analysis and processing can be facilitated, and the invention realizes the optimization of the data acquisition.

The above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the invention and are intended to be included within the scope of the invention.

Claims (8)

1. The intelligent factory data optimization acquisition method based on digital twinning is characterized by comprising the following steps of:

acquiring a historical monitoring data set corresponding to target plant equipment for digital twinning, current monitoring data and a historical cluster set corresponding to the historical monitoring data set;

performing discrete analysis processing on each history cluster in the history cluster set to obtain a discrete distribution index corresponding to the history cluster;

determining subordinate similarity indexes between the current monitoring data and each historical cluster;

determining the reference value corresponding to each parameter composing the monitoring data according to the historical monitoring data set and the current monitoring data;

Determining a target membership degree between the current monitoring data and the historical cluster according to a discrete distribution index corresponding to each historical cluster, a subordinate similarity index between the current monitoring data and the historical cluster, reference values corresponding to various parameters constituting the monitoring data, the historical cluster and the current monitoring data;

and classifying the current monitoring data and the history cluster clusters according to the target membership degree between the current monitoring data and each history cluster in the history cluster set to obtain a target cluster set so as to realize the optimization of data acquisition.

2. The intelligent factory data optimization collection method based on digital twinning according to claim 1, wherein the discrete analysis processing is performed on each history cluster in the history cluster set to obtain a discrete distribution index corresponding to the history cluster, and the method comprises the following steps:

determining a target position corresponding to each historical monitoring data according to normalized values corresponding to various parameters included in each historical monitoring data;

screening a core monitoring data set from the historical cluster according to a target position corresponding to the historical monitoring data, wherein the number of the historical monitoring data in a range surrounded by a preset radius corresponding to the core monitoring data is larger than the preset number;

Determining an aggregation position corresponding to the history cluster according to the target position corresponding to each core monitoring data in the core monitoring data set;

the distance between the target position corresponding to each history monitoring data in the history cluster and the gathering position is used as the target distance corresponding to the history monitoring data, and a target distance set corresponding to the history cluster is obtained;

and determining standard deviations of all target distances in the target distance set corresponding to the history cluster as discrete distribution indexes corresponding to the history cluster.

3. The method of claim 2, wherein determining a subordinate similarity index between the current monitored data and each historical cluster comprises:

determining a current position corresponding to the current monitoring data according to normalized values corresponding to various parameters included in the current monitoring data;

connecting the current position with the aggregation position to obtain a reference line segment between the current monitoring data and the historical cluster;

determining a vertical line between a target position corresponding to each historical monitoring data in the historical cluster and the reference line segment as a target vertical line corresponding to the historical monitoring data;

Screening historical monitoring data of the intersection of the target vertical line and the reference line segment from the historical cluster as reference monitoring data to obtain a reference monitoring data set between the current monitoring data and the historical cluster;

connecting the gathering position with a target position corresponding to each piece of reference monitoring data in the reference monitoring data set to obtain a target line segment corresponding to the reference monitoring data;

determining a first subordinate index between the current monitoring data and the historical cluster according to the reference line segment and a target line segment corresponding to each reference monitoring data in the reference monitoring data set;

determining the number of reference monitoring data in the reference monitoring data set, and the ratio of the number of historical monitoring data in the historical cluster as a second subordinate index between the current monitoring data and the historical cluster;

and determining subordinate similarity indexes between the current monitoring data and the history cluster according to the first subordinate indexes and the second subordinate indexes, wherein the first subordinate indexes and the second subordinate indexes are positively correlated with the subordinate similarity indexes.

4. A method of intelligent plant data optimization collection based on digital twinning as claimed in claim 3, wherein said determining a first subordinate indicator between the current monitored data and the historical cluster based on the target line segments corresponding to each of the reference monitored data in the reference monitored data set comprises:

determining the projection of a target line segment corresponding to each reference monitoring data in the reference monitoring data set on the reference line segment as the target projection corresponding to the reference monitoring data;

and determining a first subordinate index between the current monitoring data and the historical cluster according to target projections corresponding to each reference monitoring data in the reference monitoring data set, wherein the target projections and the first subordinate index are positively correlated.

5. The method for optimized collection of intelligent plant data based on digital twinning according to claim 1, wherein said determining a reference value corresponding to each parameter constituting the monitoring data from the historical monitoring data set and the current monitoring data comprises:

screening a candidate monitoring data set from a set formed by the current monitoring data and the historical monitoring data set according to the corresponding acquisition time of the historical monitoring data;

Determining a normalized value corresponding to the parameter included in each candidate monitoring data in the candidate monitoring data set as a target parameter value corresponding to the parameter included in the candidate monitoring data, so as to obtain a target parameter value set corresponding to the parameter;

determining the average value of all target parameter values in the target parameter value set as a representative parameter value corresponding to the parameter;

determining the absolute value of the difference value between the target parameter value corresponding to the parameter and the representative parameter value included in each candidate monitoring data in the candidate monitoring data set as a parameter difference index corresponding to the parameter included in the candidate monitoring data;

determining a difference value between the current time and the acquisition time corresponding to each candidate monitoring data in the candidate monitoring data set as a time span corresponding to the candidate monitoring data;

and determining a reference value corresponding to the parameter according to the parameter difference index corresponding to the parameter and the time span corresponding to each candidate monitoring data included in each candidate monitoring data in the candidate monitoring data set, wherein the time span and the parameter difference index are in negative correlation with the reference value.

6. The method of claim 1, wherein determining the target membership between the current monitored data and the historical cluster comprises:

determining a first membership degree between the current monitoring data and the historical cluster according to the discrete distribution index and the subordinate similarity index, wherein the discrete distribution index is in negative correlation with the first membership degree, and the subordinate similarity index is in positive correlation with the first membership degree;

determining a second membership degree of the current monitoring data and the historical monitoring data corresponding to the parameters according to the reference value corresponding to each parameter composing the monitoring data, the normalized value corresponding to the parameter included in each historical monitoring data in the historical cluster and the normalized value corresponding to the parameter included in the current monitoring data;

determining a third membership degree between the current monitoring data and the history clustering cluster according to the second membership degree of each history monitoring data in the current monitoring data and the history clustering cluster corresponding to each parameter, wherein the second membership degree and the third membership degree are positively correlated;

And determining the target membership degree between the current monitoring data and the history cluster according to the first membership degree and the third membership degree, wherein the first membership degree and the third membership degree are positively correlated with the target membership degree.

7. The method according to claim 6, wherein determining the second membership of the current monitoring data and the historical monitoring data corresponding to the parameters according to the reference value corresponding to each parameter constituting the monitoring data, the normalized value corresponding to the parameter included in each historical monitoring data in the historical cluster, and the normalized value corresponding to the parameter included in the current monitoring data comprises:

determining the absolute value of the difference value between the normalized value corresponding to the parameter included in the historical monitoring data and the normalized value corresponding to the parameter included in the current monitoring data as the data difference between the current monitoring data and the historical monitoring data corresponding to the parameter;

and determining a second membership degree of the current monitoring data and the historical monitoring data corresponding to the parameters according to the reference value and the data difference corresponding to the parameters composing the monitoring data, wherein the reference value and the second membership degree are positively correlated, and the data difference and the second membership degree are negatively correlated.

8. The method for optimized collection of intelligent plant data based on digital twinning according to claim 1, wherein the classifying the current monitoring data and the historical cluster clusters according to the target membership degree between each historical cluster in the current monitoring data and the historical cluster set to obtain a target cluster set comprises:

screening a history cluster with the maximum target membership degree from the history cluster set to be used as a candidate cluster;

determining historical clusters except the candidate clusters in the historical cluster set as target clusters;

adding the current monitoring data to the candidate cluster, and determining the added candidate cluster as a target cluster;

and forming a target cluster set by all the obtained target clusters.